different aggregation behaviour with missing values

[#1]

Hi everybody. I have the following simple data model:

set NullInterpret ='';

tab:

LOAD * INLINE [

     A,  B

    10, 20

];

tbc:

LOAD * INLINE [

     B,  C,  E

    20, 31, 50

];

tcd:

LOAD * INLINE [

    C,  D

   31, 40

     , 42

];

As you can see there is a missing value of C in the tcd table which is converted to null using NullInterpret .

The problem arises when I create the following charts:

What I don't understand is why the aggregation on A does not display the 42 value of dimension D (which seems correct to me) while the aggregation on E displays it.

The only difference I see is that on A we have only 2 tables involved, while on E we have all the 3 tables involved.

Can someone explain me?

Thanks in advance,

Filippo